|Publication number||US7016606 B2|
|Application number||US 10/204,779|
|Publication date||Mar 21, 2006|
|Filing date||Feb 28, 2001|
|Priority date||Feb 28, 2000|
|Also published as||US20030020983, WO2001065735A1|
|Publication number||10204779, 204779, PCT/2001/6218, PCT/US/1/006218, PCT/US/1/06218, PCT/US/2001/006218, PCT/US/2001/06218, PCT/US1/006218, PCT/US1/06218, PCT/US1006218, PCT/US106218, PCT/US2001/006218, PCT/US2001/06218, PCT/US2001006218, PCT/US200106218, US 7016606 B2, US 7016606B2, US-B2-7016606, US7016606 B2, US7016606B2|
|Inventors||Yi Cai, Tulay Adali, Curtis R. Menyuk|
|Original Assignee||University Of Maryland Baltimore County|
|Export Citation||BiBTeX, EndNote, RefMan|
|Patent Citations (3), Referenced by (18), Classifications (16), Legal Events (6)|
|External Links: USPTO, USPTO Assignment, Espacenet|
This application claims priority to provisional application Ser. No. 06/185,400, filed Feb. 28, 2000, entitled BLOCK SLIDING WINDOW CRITERION CODES, the contents of which are incorporated herein by reference in their entirety.
This invention is directed generally to high capacity fiber-optic communications systems and networks, and is particularly directed to mitigating optical pulse collision induced errors in fiber-optic communications systems through use of line coding.
The growth in demand for broadband services has led to increased activity in research and development of high capacity optical systems and networks. It has been predicted that quasi-linear optical communication systems such as Chirped Return-to-Zero (CRZ) and Dispersion Managed Soliton (DMS) systems will play an important role in these future networks.
The capacity of existing standard optical fiber currently installed is far greater than the capacity presently utilized. This capacity potential can, for example, be exploited through use of Wavelength-Division Multiplexing (WDM) in quasi-linear optical systems.
In optical communication systems, the main sources of errors include chromatic dispersion, fiber non-linearities, Polarization Mode Dispersion (PMD), and Amplified Spontaneous Emission (ASE) noise from the amplifiers. In WDM optical communication systems nonlinear interactions of optical pulses among different channels can cause severe inter-channel interference which may be converted into Timing Jitter (TJ) thus effectively decreasing the actual capacity.
Several different approaches such as fiber dispersion management, a jitter-tracking demultiplexer, optical equalization, Forward Error Correction (FEC), and line coding in optical systems have been proposed to combat these impairments. It has been especially shown that dramatic improvements can be obtained in repeater-less undersea systems by the use of FEC. These studies, however, are primarily based on standard FEC and line coding schemes and there has been little effort to optimize the choice of codes and to design new codes which take into account the physical mechanisms behind the impairments. Moreover, while the use of standard line codes has been studied, they have yet to be implemented in commercial systems.
With reference to
Timing Jitter (TJ) is a main source of errors and limits both a bit rate and a transmission distance in quasi-linear optical transmissions. In Wavelength-Division Multiplexing (WDM) communications, TJ also limits channel spacing and thus system capacity. As is understood by those skilled in the art, TJ is defined as the standard deviation of the timing shifts, as measured from the center of the respective time slots, of the arrival times of the sequence of optical pulses at the receiver. It is produced by several effects, including the Gordon-Haus (GH) effect, Soliton-Soliton Collisions (SSC), the acoustic effect, and Polarization Mode Dispersion (PMD).
In quasi-linear optical transmissions, periodic amplification is needed to maintain the energy of the optical pulses. Because of high gain, low insertion loss, low noise, and polarization insensitivity, Erbium-Doped Fiber Amplifiers (EDFA) are of great interest in quasi-linear transmission systems. The Amplified Spontaneous Emission (ASE) noise of the EDFA is the dominant noise source and results in not only a change of amplitudes but also carrier frequencies of optical pulses. The former causes fluctuation in the optical pulse intensity and the latter causes a fluctuation of the optical pulse arrival time (δt), which is known as the Gordon-Haus (GH) jitter. If the amplifier's bandwidth is greater than the spectral width of the optical pulses, its noise contribution can be considered as effectively white noise. Hence, a probability density function σt of the random time δt can be expressed using a Gaussian approximation with a variance term given by:
where α is a fiber attenuation coefficient; na is an amplifier's excess noise parameter; F(G) is a noise penalty function; G is an amplifier gain; z is a transmission distance; D is a fiber dispersion parameter; tFWHM is a full width at a half magnitude of an initial optical pulse; and AEff is an effective fiber core area.
Soliton-Soliton Collision (SSC) is a result of collisions among solitons that belong to different channels because of different group velocities in Wavelength-Division Multiplexing (WDM) systems. To understand the physical mechanism of SSC, consider the Non-Linear Schrödinger wave equation (NLS) for a wave traveling at a velocity u as shown in Eq. (2.1).
where i is an imaginary number i=√−1, u is the intensity, T is a constant based on physical properties of the fiber, and p(z) is a normalized group velocity dispersion profile. Because of a dependence of the second term on the distance z, Eq. 2.1 is no longer a standard NLS. However, it can be transformed into a perturbed NLS by defining:
u′=u exp(−Γz/2) and
In the transformed variables, Eq. 2.1 becomes:
The effect of SSC on the performance of WDM systems can be obtained from the NLS by considering the simplest case of two WDM channels.
In a fiber with uniform parameters, the two solitons with angular frequencies ±Ω undergo a velocity shift δΩ during a time t over a distance z defined by:
The velocity shift and its derivative (the acceleration) are shown in
1. Timing shift for each collision:
where τ is the Full Width at Half Maximum (FWHM) of a soliton pulse.
2. Collision length:
where D is fiber chromatic dispersion and Δλ is a wavelength difference of the two channels.
3. Maximum number of collisions for each soliton during the whole transmission path:
where Z is the transmission distance and T is the data transmission period.
In realistic optical WDM systems, however, the use of lumped amplifiers and fiber dispersion management has the potential to unbalance the SSC and thus cause partial SSC. Consider the situation shown in
SSC induced timing jitter is highly correlated from pulse to pulse. The net time displacement of a given pulse, from collisions with pulses of another channel, is proportional to the number of collisions that the pulse experiences as it traverses the system and that number can change by only ±1 collision from one pulse to the next. Hence, the number of bit errors caused by SSC have bursty characteristics, such as shown in
The Acoustic Effect
The acoustic effect is generated by the intensity gradient of the optical solitons transverse to the propagation direction in the optical fiber. By affecting the fiber refractive index, the acoustic effect can produce changes in central frequencies and temporal locations of the solitons and thus cause Timing Jitter (TJ). As noted previously, TJ is defined as the standard deviation of the timing shift of the arrival times of the solitons. It has been shown that the time shifts caused by the acoustic effect are Gaussian distributed. The standard deviation a of the TJ is given by the equation:
for an unfiltered soliton system, where D is fiber dispersion in psec/nm-km; z is the propagation distance in Mega-meters (Mm); τ is the FWHM of a soliton pulse; and F is the bit rate in Gbps. With guiding filters, the TJ is reduced by a factor 2/βz where β is a frequency-damping coefficient. In typical optical fibers, the acoustic waves can survive for up to approximately 100 ns and hence generate inter-symbol interference (ISI) within this time scale.
A typical normalized differentiation of the acoustic effect response function, i.e., an acoustic wave curve, is plotted in
where α is a variable incorporating a normalization of the perturbation index; N is the number of bits that fall in the duration of the acoustic wave, and T is the duration of the time slot. For one bit, the value of 1/T equals the frequency. Equation 2.8 shows that the impact of the acoustic effect on any particular bit depends in a completely deterministic way on the preceding N bits. Hence, the time shifts of solitons and the induced bit errors are highly correlated from bit to bit.
Thus, it can be seen that induced jitter in a fiber optic transmission system can have a deleterious effect and the mitigating of the errors caused thereby in a dispersion managed soliton system is highly desirable.
It is an object of the present invention to solve the above described deficiencies in prior art systems. It is a further object to attend to these deficiencies by developing an effective coding solution for mitigating major impairments in the quasi-linear systems.
Prior research has indicated that highly efficient and effective codes and filtering approaches can be developed when the physical characteristics of the optical fiber transmission line are taken into account. Thus, it is a further object of the invention to develop a coding solution based on a clear understanding of the physical mechanisms that limit the attainable data rates and distances in quasi-linear systems. Studies of the acoustic effect in single channel systems and Soliton-Soliton Collision (SSC) in Wavelength Division Multiplexing (WDM) systems have resulted in simplified models in constant dispersion fibers. Additionally, as data rates and the number of WDM channels have increased, dispersion management has become increasingly common in terrestrial optical communication systems. However, polarization effects, particularly Polarization Mode Dispersion (PMD) in terrestrial systems, and Amplified Spontaneous Emission (ASE) have generally not been considered. Hence, it is a further object of the invention to develop the coding solution in consideration of the effects of dispersion management, particularly the PMD and ASE as they relate to the bit error patterns in quasi-linear systems.
It is a further object of the invention to develop effective coding methods based on the basic concept of a Sliding Window Criterion (SWC) and block SWC codes. Concatenated Sliding Window Criterion/Reed-Solomon code (SWC/RS) is an efficient and effective solution for reducing collision-induced timing jitter in optical soliton systems. Thus, it is a further object of the invention to define the relationship between the parameters of the SWC code and the quasi-linear WDM systems to develop rules underlying design of optimal SWC codes for given WDM system parameters. This includes trellis-based SWC coding schemes because of the natural match between a sliding window nature of the physical effects in optical communications and the operation of trellis-based encoding. It is a further object of the invention to develop an adaptive digital approach to compensate the data pattern dependent and highly correlated bit errors in quasi-linear systems.
Finally, it is a further object of the invention to be able to verify the developed coding methods.
These and other features and advantages of the present invention will be understood by those skilled in the art by reference to the following detailed description and appended drawings wherein like numerals represent like elements through the several views.
The present invention utilizes the concept of a Sliding Window Criterion (SWC) to develop a line coding method for a computer memory in an optical Wavelength Division Multiplexed (WDM) communications. SWC is defined as a metric that takes into account physical mechanism of errors in optical communications. In particular the present invention utilizes an SWC-based design approach for mitigating errors in quasi-linear transmission schemes.
The main motivation behind the coding scheme of the present invention can be explained by considering the simplified model of Soliton-Soliton Collisions (SSC) such that all collisions are complete collisions. However, the present invention is not limited to soliton effects and can be used to mitigate other errors in a WDM communications system. If a two channel case with optical frequency difference Δf is initially considered, a simplified model of Soliton-Soliton Collisions (SSC) can be described by equations 2.4 to 2.6.
When there are complete collisions, after each collision, the faster of the colliding solitons is advanced and the slower one is delayed by the same absolute value of arrival time shift δt. Given a plurality of system parameters, including Z (transmission distance), D (fiber chromatic dispersion), Δf (change in frequency), T(transmission time slot width), and τ (Full Width Half Maximum of a soliton pulse), the collision length Lcoll, defined as the length between the beginning and end points where the solitons overlay at their half power points, can be calculated. The number of collisions each soliton experiences, Nch1ch2, can also be calculated if data sequences of all marks (i.e. all “1”s) are transmitted in both channels. The total timing shift introduced by SSC of each soliton after traversing the whole transmission path is thus simply the product of the number of collisions N that the soliton experiences and the timing shift δt for each collision. Thus, it is straightforward to obtain an equation for the timing shift in a Wavelength Division Multiplexing (WDM) system with more than two channels by using the above equations for each pair of channels and summing the results over all channels.
Since the timing shift δt for each collision is constant for a given value of τ and Δf, a total timing shift of each soliton only depends on a value of N, the number of collisions, which is determined by a pattern of transmitted data in other channels. Given that Timing Jitter (TJ) is nothing but the deviation of the timing shifts of transmitted solitons, the TJ can be effectively decreased if every soliton experiences almost the same number of collisions throughout the entire transmission path. This can be seen, for example, with reference to
Since in the example of
Since the overall effect is seen in sliding blocks 518, 520, and 522, a random variable K is defined as a number of marks (e.g. where a mark is a “1” in a data stream) in a sliding window. The criteria of a sliding window includes a performance index. The performance index of a Sliding Window Criteria or SWC over a block of length “L” is defined as: SWCL=var (K), quantifies the above consideration in that minimization of the variance term will lead to every soliton experiencing a similar number of collisions. The SWC of the present invention is used to construct a mapping code whereby a block of a binary input data sequence is mapped to a corresponding block of encoded data whereby the variance of the number of possible soliton-soliton collisions is reduced in comparison to the original input data.
Referring now to
Converter 616 is connected to an SWC code encoder memory 616 which has 2N memory addresses, such as address 616 a. Each of the 2N memory addresses 616 a corresponds to a unique combination of N bits in N-bit block 614. For each memory address 616 a there is a one-to-one mapping to a corresponding encoded data 616 b having N plus M bits. In the example of
The output of memory 616 is connected to an input address port of a parallel-to-serial converter 620. Converter 620 converts the parallel output of memory 616 to a serial data stream 622. Like converter 612, converter 620 can be either a hardware implementation, such as a conventional parallel-in serial-out asynchronous or synchronous shift register, or a software implementation. Converter 620 is connected to a conventional optical modulator 624 which converts an input electrical pulse sequence to an optical pulse sequence 626.
The output of optical modulator 624 is connected to a transmitter end of a conventional optical fiber channel 628. The receiver end of optical fiber channel 628 is connected to a conventional optical detector that receives a pulse sequence 630 from fiber channel 628. Optical detector 632 converts pulse sequence 630 to serial data sequence 634 of electrical signals. Data sequence 634 is converted by a conventional serial-to-parallel converter 636 to 10-bit blocks 638. Converter 636 can be similar to converter 612, but instead uses a 10-bit serial word.
The output from converter 636 is connected to the address input of an SWC code decoder memory 640. Memory 640 has an inverse mapping of 2N memory addresses, such as address 640 a, to data, such as shown at 640 b, corresponding to the mapping of encoder 616. Each of the 2N memory addresses 640 a corresponds to a unique data 616 b of SWC code encoder 616. For each memory address 640 a there is a one-to-one mapping to a corresponding decoded data 640 b. A 10-bit block from the received data sequence is provided to SWC code decoder memory 640 as an address signal 640 a, each 10-bit address pointing to a memory location in which an 8-bit decoded data signal 640 b is stored and which replicates an originally provided 8-bit block 614. This data stored in the addressed memory location is provided as a parallel output of 8 bits from SWC code decoder memory 640.
The output of decoder memory 640 is connected to a parallel-to-serial converter 644. Converter 644 converts the parallel output from decoder memory 640 to a serial data stream 646 which is a replicate of input data sequence 610. Converter 644 is similar to, or can be the same as converter 620.
As discussed above, the encoding of an input data word into a block SWC codeword can, by way of example and not by way of limitation, be implemented by writing a mapping table into a memory chip and using an input data block as a memory address. Thus an output of the memory chip is just the encoded data mapped to a corresponding memory address location. An encoding speed is determined by a read cycle time of the memory chip. Similarly, encoded data can be received as the memory address to implement decoding. Currently, a number of high-speed memory chips are commercially available. For example, the Motorola MCM64E918 RAM chip with 19-bit address and 18-bit output to implement a 16B18B SWC code. A minimum read cycle time that can be achieved with this chip is 3 ns, hence an 18 bit/3 ns can be achieved, i.e., 6 Gbps encoding and decoding speeds are achieved. By using k of these chips in parallel, as high as 6 k Gigabits per second (Gbps) encoding and decoding speeds can be achieved.
A first block 730 of 8-bits (thus N=8) of input stream 710 is seen to have all “1”s. Block 730 corresponds to memory address 716 a n in SWC code encoder 616. Mapped to address 716 a n is data codeword 716 n. Data codeword 716 b n corresponds to a first encoded 10-bit block 740. A second block 732 of 8-bits of input stream 710 is seen to have two “0”s, four “1”s and two “0”s. Block 732 corresponds to memory address 716 a i in SWC code encoder 616. Corresponding to address 716 a i is data codeword 716 b i. Data codeword 716 b i corresponds to a second encoded 10-bit block 742. A third block 734 of 8-bits of input stream 710 is seen to have all “0”s. Block 734 corresponds to memory address 716 a 1 in SWC code encoder 616. Corresponding to address 716 a 1 is data codeword 716 b 1. Data codeword 716 b 1 corresponds to a third encoded block 744.
The development of the code words stored in memories 616 and 640 will now be discussed. Based on the SWC, a block coding or line code approach which when concatenated with a Reed-Solomon (RS) code provides very effective mitigation of errors. However, when using block codes the ends of the code words need to be considered as well. Thus, to help in the development of code word selection, consider the following definitions:
Fragmental: An n-bit binary block is defined as fragmental if it has at least one transition, i.e., there is at least one occurrence of either a “1” bit followed by a “0” bit or a “0” bit followed by a “1” bit in the n-bit block. A binary signal sequence is defined as n-bit fragmental if any n-bit block in the sequence is fragmental.
Fragmentation degree (FD): An n-bit fragmentation degree of a binary code word is defined as: FDn=ml(l−n+1) where FDn ε[0,1], l is a length of a code word, and m is a number of n-bit fragmental blocks in the code word.
Fragmental end (FE): A binary code word is defined as having n-bit fragmental ends (logical TRUE) if its first n bits and last n bits are n-bit fragmental.
The construction of a SWC data-code word mapping table of the present invention determines the ultimate performance of the code. In a preferred mapping table, in the SWC sense, i.e., a mapping table that minimizes an SWC, both a length of a code word and a length of a sliding window are taken into consideration. Hence, code words are selected with: (1) a same number of marks; (2) high fragmentation degrees; and (3) fragmental ends. If the SWC code word is much shorter than the sliding window, there can be several code words within the sliding window implying heavier dependency on the number of marks in the code words than on their FDs. Hence, in a first preferred embodiment of the present invention, rule (1) of code word selection is more heavily weighted as compared to rule (2). Conversely, if the SWC code word is longer than the sliding window, there is less than one code word within the sliding window. Hence SWC will depend more on the FDs of the code words than the numbers of marks in the code words. In this case, in a second preferred embodiment of the present invention, rule (2) is more heavily emphasized than rule (1).
Based on these observations, two alternative mapping table generation algorithms have been developed depending on the emphasis, the Fragmentation-First (FF) algorithm, and the Equal-1-First (E1F) algorithm.
Flow diagrams of the FF and E1F algorithms are shown in
With reference to
With reference to
In the example of
For a random binary input sequence with equal probability of “1”s and “0”s in the sequence (p=0.5), all code words have the same mapping probability and hence it doesn't matter how the code words are arranged in the mapping table. However, if the probability of “1”s and “0”s in the input sequence are unequal, better performance of SWC can be achieved by assigning code words with better SWC features (i.e., lower variance) to input codes with high probabilities. Thus, the mapping tables are divided into several parts according to the pattern of the selected code words.
The influences of the two algorithms of the present invention on the Power Spectral Density (PSD) of a transmitted signal evaluated using a spectral analysis approach in comparison with two conventional encoding schemes is shown in
As expected, PSD plot 902 of Fragmentation-First algorithm 800 exhibits larger components at high frequencies and hence implies higher transition density than PSD plot 904 of Equal-1-First algorithm 850. However, PSD plot 904 of Equal-1-First algorithm 850, shown by its smaller components at low frequencies, is more balanced than PSD plot 902 of Fragmentation-First algorithm 850.
In an ideal system, i.e., one with no soliton-soliton-collisions and hence no SSC induced timing shifts, each soliton pulse is centered in its respective time slot. However, where soliton-soliton collisions occur, the timing shift caused by the SSC displaces the various soliton pulses from the centers of the time slots. If, for example, a pulse is displaced by ±0.5T, the displaced pulse is then centered at a boundary between successive time slots Ti and Ti+1. Similarly, if a pulse is displaced by greater than +0.5T, the displaced pulse is then centered in the next successive time slot; and if the pulse is displaced by greater than −0.5T, the displaced pulse is centered in the preceding time slot. Thus, it is possible to misconstrue a pulse which is centered in a time slot as belonging to that time slot when, in fact, the pulse could belong to another time slot. To reduce the likelihood of associating a soliton pulse with an incorrect time slot, an “acceptance window” having a width less than the width of the time slot, but centered in the time slot, is defined. A pulse is construed as being associated with a particular time slot if the pulse is centered in the acceptance window. Thus, an acceptance window in which the ratio of the width of the acceptance window to the width of the time slot approaches “1” in value is less discriminating whereas an acceptance window in which the ratio of the width of the acceptance window to the width of the time slot is smaller is more discriminating.
With reference to
Plot 1002 represents a probability distribution of a soliton time shift for a binary data stream without coding. Plot 1004 represents a probability distribution of a soliton time shift for a binary data stream with Reed-Solomon (RS) coding; and plot 1006 represents a probability distribution of a soliton time shift for binary data stream with concatenated Sliding Window Criterion/Reed Solomon (SWC/RS) coding of the present invention. As can be seen in
Simulation results are presented for some selected data patterns to demonstrate the effectiveness of the coding scheme according to the present invention.
With reference to
|Cited Patent||Filing date||Publication date||Applicant||Title|
|US5638070 *||Aug 15, 1995||Jun 10, 1997||Victor Company Of Japan, Ltd.||Signal processing method and apparatus for transforming N-bit codes to M-bit codes wherein M is greater than N|
|US6079007 *||Mar 10, 1998||Jun 20, 2000||Pan Atlantic Corporation||System and method for generating pseudo-random codes|
|US6308249 *||May 14, 1999||Oct 23, 2001||International Business Machines Corporation||Storing data in a grey code system|
|Citing Patent||Filing date||Publication date||Applicant||Title|
|US7254331 *||Aug 9, 2002||Aug 7, 2007||Micron Technology, Inc.||System and method for multiple bit optical data transmission in memory systems|
|US7313332 *||Apr 21, 2004||Dec 25, 2007||Stmicroelectronics S.R.L.||Method of transmitting data streams on optical links and system and computer program product therefor|
|US7746095||Jun 8, 2009||Jun 29, 2010||Round Rock Research, Llc||Memory module and method having improved signal routing topology|
|US7805586||May 10, 2006||Sep 28, 2010||Micron Technology, Inc.||System and method for optimizing interconnections of memory devices in a multichip module|
|US7836252||Aug 29, 2002||Nov 16, 2010||Micron Technology, Inc.||System and method for optimizing interconnections of memory devices in a multichip module|
|US7870329||May 3, 2006||Jan 11, 2011||Micron Technology, Inc.||System and method for optimizing interconnections of components in a multichip memory module|
|US7899969||Oct 15, 2009||Mar 1, 2011||Round Rock Research, Llc||System and method for memory hub-based expansion bus|
|US7949803||Aug 31, 2009||May 24, 2011||Micron Technology, Inc.||System and method for transmitting data packets in a computer system having a memory hub architecture|
|US20040223767 *||Apr 21, 2004||Nov 11, 2004||Francesco Pappalardo||Method of transmitting data streams on optical links and system and computer program product therefor|
|US20040251929 *||Jun 11, 2003||Dec 16, 2004||Pax George E.||Memory module and method having improved signal routing topology|
|US20050030797 *||Sep 1, 2004||Feb 10, 2005||Pax George E.||Memory module and method having improved signal routing topology|
|US20050091464 *||Oct 27, 2003||Apr 28, 2005||Ralph James||System and method for using a learning sequence to establish communications on a high-speed nonsynchronous interface in the absence of clock forwarding|
|US20050146946 *||Feb 18, 2005||Jul 7, 2005||Taylor George R.||System and method for optically interconnecting memory devices|
|US20050210216 *||Mar 8, 2004||Sep 22, 2005||Jobs Jeffrey R||Memory hub architecture having programmable lane widths|
|US20050216648 *||Mar 25, 2004||Sep 29, 2005||Jeddeloh Joseph M||System and method for memory hub-based expansion bus|
|US20050228939 *||Apr 8, 2004||Oct 13, 2005||Janzen Jeffery W||System and method for optimizing interconnections of components in a multichip memory module|
|US20060023528 *||Sep 28, 2005||Feb 2, 2006||Pax George E||Memory module and method having improved signal routing topology|
|US20110299316 *||Dec 8, 2011||Choi Jung-Hwan||Memory module, method and memory system having the memory module|
|U.S. Classification||398/27, 398/25, 398/159, 714/701, 398/39, 711/217|
|International Classification||H04B10/08, H04K3/00, G06K5/04, H04B10/18, H04J14/02, G06F12/00|
|Cooperative Classification||H04B10/25077, H04J14/02|
|European Classification||H04B10/25077, H04J14/02|
|Aug 23, 2002||AS||Assignment|
Owner name: UNIVERSITY OF MARYLAND BALTIMORE COUNTY, MARYLAND
Free format text: ASSIGNMENT OF ASSIGNORS INTEREST;ASSIGNORS:CAI, YI;ADALI, TULAY;MENYUK, CURTIS R.;REEL/FRAME:013419/0843;SIGNING DATES FROM 20020808 TO 20020812
|Apr 25, 2006||AS||Assignment|
Owner name: AIR FORCE, UNITED STATES, VIRGINIA
Free format text: CONFIRMATORY LICENSE;ASSIGNOR:UNIVERSITY OF MARYLAND BALT CO CAMPUS F49620-96-1-0439;REEL/FRAME:017527/0787
Effective date: 20060327
|Sep 11, 2009||FPAY||Fee payment|
Year of fee payment: 4
|Nov 1, 2013||REMI||Maintenance fee reminder mailed|
|Mar 21, 2014||LAPS||Lapse for failure to pay maintenance fees|
|May 13, 2014||FP||Expired due to failure to pay maintenance fee|
Effective date: 20140321